Training now works / bug memory error while training

Assets/Brains/CozmoLearningBrain.asset

@@ -15,14 +15,13 @@ MonoBehaviour:
   brainParameters:
     vectorObservationSize: 0
     numStackedVectorObservations: 1
-    vectorActionSize: 0200000002000000
+    vectorActionSize: 04000000
     cameraResolutions:
     - width: 320
       height: 240
       blackAndWhite: 0
     vectorActionDescriptions:
-    - forward
-    - rotation
+    - 
     vectorActionSpaceType: 0
   model: {fileID: 0}
   inferenceDevice: 0

Assets/Scripts/ML Cozmo/CozmoAgent.cs

@@ -51,10 +51,12 @@ public class CozmoAgent : Agent
     // to be implemented by the developer
     public override void AgentAction(float[] vectorAction, string textAction)
     {
-
+        //print("Action before FloorToInt: " + vectorAction[0]);
         int action = Mathf.FloorToInt(vectorAction[0]);
         Point centerOfGravity = onRenderImageTest.CenterOfGravity;
-        Vector3 targetPos = transform.position;
+        //Vector3 targetPos = transform.position;
+
+        //print("Action after FloorToInt: " + action);
 
         AddReward(-0.01f);
 
@@ -73,7 +75,7 @@ public class CozmoAgent : Agent
                 movement.Turn(-1);
                 break;
             default:
-                movement.Move(0);
+                //movement.Move(0);
                 throw new ArgumentException("Invalid action value. Stop movement.");
         }
 
@@ -87,21 +89,18 @@ public class CozmoAgent : Agent
         // If centerOfGravity lies near to the center of the image horizontally
         if (centerOfGravity.X > renderCamera.targetTexture.width / 2 - nearAreaLimit && centerOfGravity.X < renderCamera.targetTexture.width / 2 + nearAreaLimit)
         {
-            Done();
             SetReward(1);
             print("Reward: +1");
         }
         else if (centerOfGravity.X > renderCamera.targetTexture.width / 2 - farAreaLimit && centerOfGravity.X < renderCamera.targetTexture.width / 2 + farAreaLimit)
         {
-            Done();
-            SetReward(-1);
-            print("Reward: -1");
+            SetReward(-0.7f);
+            print("Reward: -0.7");
         }
         else
         {
-            Done();
-            SetReward(-2);
-            print("Reward: -2");
+            SetReward(-1);
+            print("Reward: -1");
         }
     }